Ultrafiltration UF membranes can retain macromolecular solutes. Solute retention is mainly determined by solute size.

However, other factors such as solute-solute and solutemembrane interactions can affect solute retention. Ultrafiltration is used for: 1. Concentration of solutes 2. Purification of solvents 3. Fractionation of solutes 4. Clarification UF is attractive because of the high throughput of product, low process cost and ease of scale-up. UF is now widely used for processing therapeutic drugs, enzymes, hormones, vaccines, blood products and antibodies. The major areas of application are listed below: 1. Purification of proteins and nucleic acids 2. Concentration of macromolecules 3. Desalting, i.e. removal or salts and other low molecular weight compounds from solution of macromolecules 4. Virus removal from therapeutic products The pore size as related to molecular weight cut-off (MWCO) is shown in the table:

Ultrafiltration separates solutes in the molecular weight range of 5 kDa to 500 kDa. UF membranes have pores ranging from 1 to 20 nm in diameter. Most UF membranes are anisotropic, with a thin "skin layer", typically around 10 urn thick fused on top of a microporous backing layer. The skin layer confers selectivity to the membrane while the microporous backing layer provides mechanical support. The ability of an ultrafiltration membrane to retain macromolecules is traditionally specified in terms of its molecular cut-off (MWCO). A MWCO value of 10 kDa means that the membrane can retain from a feed solution 90% of the molecules having molecular weight of 10 kDa. This is a highly subjective definition since it does not specify any other conditions such as feed concentration, transmembrane pressure and so on. However, the MWCO provides a good starting point for selecting a membrane for a given application. The flow of a solvent through ultrafiltration membranes can be described in terms of a pore flow model which assumes ideal cylindrical pores aligned normal to the membrane surface:

The above equation can easily be obtained using Poiseuilles law which describes the liquid flow through a pore (q) where:

The flux, or flow per unit membrane area, is the sum of all the flows through the individual pores (N pore) and so is given by:

For membranes of equal pore area and porosity (), the number of pores per square centimeter is proportional to the inverse square of the pore diameter. That is:

Substitution in the flux equation results in the overall relation presented above. Ultrafiltration is usually operated in the cross-flow mode where the feed flows parallel to the membranes surface as shown in Fig. 11.17. The dead ended mode of operation where the feed is pumped normal to the membrane surface is rarely used in ultrafiltration processes. The transmembrane pressure in cross-flow UF is given by:

The transmembrane pressure in UF is usually generated by pressurizing the feed side with a valve on the retentate line. The transmembrane pressure could also be generated by negative pressure, i.e. by providing a pump on the permeate side which would draw the permeate by suction. The first option is preferred in constant pressure ultrafiltration while the second option is used for constant flux ultrafiltration.

Diafiltration Filtration in which a solution is continuously recycled through a membrane filtration device so that the process stream containing the permeating species is removed. New solvent is added to the reaction vessel while this material is being removed, causing the reactor contents to be free of membrane-permeating species within a brief period of time. Rejection coefficient (R) It is possible to dene a solute rejection coefcient (R) by: R = 1 (Cp/Cf) where Cf is the concentration of solute in the feed stream and Cp is the concentration of solute in the permeate. For a given ultraltration membrane with a distribution of pore sizes there is a relationship between R and the solute molecular weight, as shown in the figure below.

Ion-exchange membranes, which are used for electrodialysis, usually consist of highly swollen charged gels prepared either by dispersing a conventional ion-exchange material in a polymer matrix, or from a homogenous polymer in which electrically charged groups such as sulphonic, carboxylic or quarternised amine groups have been introduced. The rst type is referred to as a heterogeneous membrane, while the second type is termed a homogeneous membrane. A membrane with xed positive charges is referred to as an anion exchange membrane since it may bind and hence selectively transport anions from the surrounding solution. Similarly, a membrane containing xed negative charges is termed a cation exchange membrane. Ionexchange membranes exclude, that is, do not bind and do not allow the transport of, ions which bear charges of the same sign as the membrane. GENERAL MEMBRANE EQUATION It is not possible at present to provide an equation, or set of equations, that allows the prediction from rst principles of the membrane permeation rate and solute rejection for a given real separation. Research aimed at providing such a prediction for model systems is under way, although the physical properties of real systems, both the membrane and the solute, are complex. An analogous situation exists for conventional ltration processes. The general membrane equation is an attempt to state the factors which may be important in determining the membrane permeation rate for pressure driven processes. This takes the form:

where J is the membrane ux, expressed as volumetric rate per unit area, |P| is the pressure difference applied across the membrane, the transmembrane pressure, is the difference in osmotic pressure across the membrane, Rm is the resistance of the membrane, and Rc is the resistance of layers deposited on the membrane, the lter cake and gel foulants. If the membrane is only exposed to pure solvent, say water, then the equation reduces to J =|P|/Rm. Knowledge of water uxes is useful for characterising new membranes and also for assessing the effectiveness of membrane cleaning procedures. In the processing of solutes, the equation above shows that the transmembrane pressure must exceed the osmotic pressure for ow to occur. It is generally assumed that the osmotic pressure of most retained solutes is likely to be negligible in the cases of microltration. The resistance Rc is due to the formation of a lter cake, the formation of a gel when the concentration of macromolecules at the membrane surface exceeds their solubility giving rise to a precipitation, or due to materials in the process feed that adsorb on the membrane surface producing an additional barrier to solvent ow. The separation of a solute by a membrane gives rise to an increased concentration of that solute at the membrane surface, an effect known as concentration polarization. This may be described in terms of an increase in . It is within the framework of this equation that the factors inuencing membrane permeation rate will be discussed in the following sections. The layers deposited on the membrane during cross-ow microltration are sometimes thought to constitute dynamically formed membranes with their own rejection and permeation characteristics.

As most ultrafiltration membranes can not be visualized as having parallel cylindrical pores, a parameter, the membrane hydraulic resistance is used for calculating permeate flux:

Where Rm = membrane hydraulic resistance (Pa s/m) When a solution of macromolecules is ultrafiltered, the retained macromolecules accumulate near the membrane surface. This is known as concentration polarization. At steady state, a stable concentration gradient exists near the membrane owing to back diffusion of solute from the membrane surface. Concentration polarization not only offers extra hydraulic resistance to the flow of solvent but also results in the development of osmotic pressure which acts against the applied transmembrane pressure. Therefore:

If the solute build-up is extensive, a gel layer may be formed on top of the membrane. In such a situation, the equation needs to be modified to account for the gel layer resistance:

The formation of the concentration polarization layer and the gel layer can be inferred from the permeate flux versus transmembrane pressure profile for an ultrafiltration process involving partially or totally retained solutes. Fig. 11.18 shows a typical UF flux - pressure profile. At lower values of transmembrane pressure, the permeate flux increases linearly with increase in pressure, almost coinciding with the solvent profile. However, as the pressure is further increased, there is deviation from the solvent profile, this being due to concentration polarization. At very high transmembrane pressures, the permeate flux usually plateaus off, clearly suggesting the formation of a gel layer. Beyond this point, increasing the transmembrane pressure has a negligible effect on the permeate flux, this value of permeate flux being referred to as the limiting flux (Jlim). The profile shown in Fig. 11.18 can be divided into two zones: the pressure dependent zone consisting of the linear part of the profile and the non-linear part due to concentration polarization in which increase in transmembrane pressure still results in some increase in permeate flux, and the pressure independent zone

where the limiting flux has been reached. Ultrafiltration processes are usually operated in the pressure dependent zone.

Due to the build-up of rejected solute molecules near the membrane surface and the resulting back diffusion of solutes into the feed, a concentration profile of solute molecules such as shown in Fig. 11.19 is obtained. The separation of process liquid and solute that takes place at the membrane during ultraltration gives rise to an increase in solute concentration close to the membrane surface, as shown in the figure below. This is termed concentration polarisation and takes place within the boundary lm generated by the applied cross-ow.